TAS Senate - Below the line voting analysis

Between work and other commitments, I have whipped together a quick graphical analysis of how preferences actually flowed below the line (BTL) in a number of interesting mid-stage eliminations in the declared Tasmanian Senate.


Firstly, I acknowledge the work done by Antony Green to put all the numbers in a tabulated form. My analysis here breaks down the vote by ideologies, excludes known Above The Line (ATL) votes, represents preference flows graphically, and provides some clues as to how preferences will flow in the close WA contest.

For purposes of easy analysis, I have assumed the following ideological breakdowns by party:

Left	Central	Right	Religious
HEMP	Australian Independents	Liberal Democrats	Family First
Pirate party	Palmer United Party	Shooters and Fishers	Democratic Labour Party
Sex Party		Stop the Greens	Australian Christians
Greens

(Apologies in advance to any FF and DLP supporters who wish to deny the party is religious - when sorting by voting patterns, there is clear evidence to suggest that these parties support other religious parties to a large extent. Accordingly, they're grouped together).

I have placed PUP in the centre not the right to acknowledge the fact that it is likely to be a party where both Labor and Liberal voters would likely prefer it over the other major party. This is in contrast to other smaller left and right parties where major party voters may, in some instances, prefer the opposing major party before other minor parties of opposite and more extreme ideology.

These graphs represent all BTL preference flows upon the elimination of each named party.

1. Stop the Greens
This elimination displayed a typical "scattergun" approach, where preferences were split almost equally. Note that the largest beneficiary of STG preferences was FF - which was coincidentally next to the STG ticket. It would be almost perfectly accurate in this instance to just model this as "all parties get an equal share".

2. Christians
This produced the strongest BTL preference flow of anything that I analysed. This is remarkable given there are approx a dozen alternative parties and more than half of all Christian BTL voters found FF on the other side of the ballot paper to stick a number "3" next to. These votes exhibited a very strong tendancy to avoid going to Left parties.

3 & 4. Hemp & Pirate
Both of these small left parties shows a very strong tendency to preference the Sex Party ahead of both the Greens and Labor. It was this strong tendency which caused me to slightly misunderestimate the eventual close 244 vote margin that separated Labor and Sex Party at a later stage. Unsurprisingly, very little vote of these parties crossed to the religious parties. A moderate percentage of HEMP vote made its way to the shooters because they were immediately to the right of the HEMP party (not because "shooting up" is a policy of either party!)

5. DLP
Just like the Christian vote previously, FF was the main beneficiary of DLP BTL votes. Greens did surprisingly well, but again they were the next continuing ticket on the ballot to the right of the DLP ticket.

6. AI
The Australian independents produced an almost perfectly random distribution, with a moderate preference to the Greens.

7. Shooters and Fishers
The main beneficiaries of the Shooters BTL vote were the Sex Party (perhaps the common link is voters for both parties like having a good time? But the Sex Party was the next continuing party on the right of the ballot) and the Palmer Party (more obvious commonality).

8. Family First
Unsurprisingly, FF BTL voters preferred Liberal than any other party. PUP also benefited

9. Sex Party
No surprises - these went strongly green, but with significant leakage to Palmer. If we consider a hypothetical GRN vs ALP contest, a Sex Party ATL vote would prefer Greens 100%, but BTLs would only give a +16% margin to the Greens.

10. ALP
This exclusion equally favours Greens and PUP, in line with previous analysis I have done that shows that there is significant ALP leakage away from the Greens. Arguably, this is caused by ALP voters happy to be preferencing the Greens instead just voting 1 above the line!

Summary
- BTL preferences scatter considerably.
- For central parties (like AI), one can safely assume they scatter randomly.
- Religious parties stick more tightly to each other than votes of other ideologies.
- PUP does fairly well from everyone
- Left BTL votes strongly favour other Left parties ahead of the Greens

Implications for modelling
In accordance with my previous work on this topic, I consider that the best way to model BTL votes is to ignore them! As long as there is a mix of parties of different ideologies, any such trends will tend to cancel each other out over a number of preference distributions.